Compressibility of filter cakes in dead-end hollow fibre ultrafiltration

Low-pressure membrane technologies are recognized by the drinking water industry as very attractive processes. In the year 2000, world-wide about 1.5 2 million cubic meters per day of drinking water is being produced using ultrafiltration. Porous membranes lose their hydraulic performance as material accumulates on their surfaces and/or within their pores (fouling). A key issue for design and operating purposes is the impact of membrane fouling on the flux reduction. However, membrane fouling is far from being fully understood, and understanding how membranes foul is the key for developing effective fouling control strategies. The addition of coagulant to the feed water has been suggested to improve permeate quality and permeate flux in UF of surface water with high organic content. Previous research shows that the coagulation efficiency may change according to raw water quality and type of coagulant used. This research aimed to contribute to a better understanding of the role of coagulant in fouling of UF membranes and intended to describe the behaviour of the filter cake in deadend UF of surface water, containing natural organic matter and particles. The values of the specific cake resistance of filter cakes build up from coagulated water at a flux of 150 and 50 l/m2.h were respectively 83 and 730/0 lower than in the case of noncoagulated water for the same filtered volume. The increase in specific cake resistance during cake built up was (almost) linear for coagulated water; indicating a hardly compressible or incompressible cake, while for non-coagulated water at a flux of 150 I/m2.h the cake was compressed, probably from the start of the filtration cycle. The reversibility of the cake compression in the experiments carried out for this research was between 90 and 1200/0 for cakes built up with non-coagulated water and between 95 and 1050/0 for cakes built up with coagulated water. The irreversible part of the compression is caused by the retardation effect and possible depth filtration. The power relationship a =aO~:ke fits the results of the constant flux UF experiments. The average n-value of the cakes built up with non-coagulated water at fluxes of 150 and 50 l/m2.h was 0.43, which indicated that cake is medium compressible. The n-values found for cakes built up with coagulated water were about 50% lower compared to cakes built up with non-coagulated water. Higher cake masses (resulting from higher flux) gave higher cake compressibility and lower initial specific cake resistance. Coagulation prior to UF during filtration gave a lower the rate of increase of the TMP over time (factor 4 to 4.6) and a slower increase of the combined membrane and blocking resistance after BW (factor 2), compared to filtering non-coagulated water. The cleaning was more effective when coagulation is applied, because aggregates are easier removedthan small particles and (non-coagulated) NOM, which can form a sticky gel layer. The effective membrane area reduced with 30% after 4 filtration cycles and with more than 40% after 8 filtration cycles at a flux of 50 IIm2.h with non-coagulated water and with 90/0 after 4 filtration cycles at a flux of 150 l/m2.h with non-coagulated water. Cleaning appears more effective at higher fluxes. It is recommended to improve the MFI-UF equipment in such a way that a really ""constant""flux can be maintained throughout the entire experiments